Restricted or Liberal Fluid Therapy

Abstract

While fluid deprivation and fluid overdose resulting from negligence or misinformed prescription are surely harmful, observations and experiments do not point to fluid therapy as a significant determinant of patient outcomes from major surgery. We can take steps to regulate (a) the extracellular fluid volume and (b) the intravascular—extravascular distribution of the extracellular fluid. On point (a), maintaining a positive input—output fluid balance of 1 to 2 liters on the day of surgery seems to be optimal for modern minimally-invasive surgery. On point (b) there are two interdependent extracellular fluid circulations, of blood and of interstitial fluid. The determinant of the equilibrium between plasma volume and interstitial volume are the transendothelial filtration rate (J_v) of fluid from the blood circulation to the interstitial circulation and the pre-nodal lymph flow rate (Q_lymph). Both are affected by anesthesia and by vasoactive agents. We can minimise post-operative edema (harmful interstitial fluid accumulation) and optimize intravascular fluid volume by taking steps to balance fluid input, fluid output, J_v and Q_lymph.

FormalPara Key Points

• 1. A two-dimensional approach to perioperative fluid therapy manages the water volume balance and the plasma volume to interstitial fluid volume ratio.

• 2. Filtration rate (J_v) of fluid from the plasma to the interstitial fluid is a major determinant of the dynamic equilibrium between plasma volume and interstitial volume. It is affected by anesthesia and by vasoactive agents.

• 3. Pre-nodal lymph flow rate (Q_lymph) is closely matched to J_v in health, and the way to optimize plasma volume to interstitial fluid volume ratio is to take steps to balance J_v and Q_lymph.

• 4. In hypovolemia, when microvascular pressures are low, spontaneous plasma volume refill is by protein-rich lymph via the efferent lymphatics and occurs at about 1 ml per minute. J_v is close to zero during hypovolaemia and so the colloid osmotic pressure of the resuscitation fluid has little effect on resuscitation volume.

• 5. Observation and research indicate that the red cell dilution efficiency of colloid solutions is up to five times that of crystalloids, but the resuscitative efficiency of colloids is only 1.5 times. Colloid-induced anemia and hypoalbuminemia in resuscitated patients is now well documented.

• 6. A false distinction is often made between situations in which capillaries are “leaky” as in sepsis, or non-leaky as in uncomplicated perioperative cases. This false distinction is used to legitimize peri-operative colloid use. In fact trauma, surgery, and sepsis are all associated with systemic inflammatory response, and there is no reason to distinguish them for the purposes of rational fluid prescribing.

• 7. New insights into non-osmotic sodium disposition probably explain the “Sodium Paradox” and the perioperative phenomenon of “Missing Sodium”.

• 8. While hypotonic solutions may be safe and rational for the maintenance of fasted subjects, free water load cannot be cleared by arginine vasopressinemic perioperative patients whose renal collecting ducts are rendered permeable to water.

Introduction

It has long been supposed that there must be an optimal approach to fluid therapy for trauma and surgery that delivers lower complication rate and lower mortality. Too little puts the patient at risk of hypoperfusion, and too much leads to edema. Francis Moore was an early contributor to the scientific investigation of this hypothesis [1, 2]. His numerous studies in the 1950s and 1960s suggested that the stress of trauma and surgery caused salt and water retention, so that it is rational to restrict salt and water administration. Tom Shires measured a reduction in the effective extracellular fluid volume after major surgery, and attributed this to the development of an isolated “third space” of extracellular fluid due to paralytic ileus and edema at the site of injury [3]. He therefore proposed that it was rational to administer liberal amounts of isotonic salt solution to maintain the effective extracellular fluid volume [4, 5]. These opposing views so polarized the practices of surgeons and anesthetists that Moore and Shires published a plea for “Moderation” in the major surgical and anesthesia journals in 1967 [6].

Moore fully appreciated that there is a steady state, a dynamic equilibrium, between the volume of fluid in the interstitial phase and that found in the plasma. This partition of the extracellular fluid he expressed as the PV:IF ratio.[ref] The PV:IF ratio is around 0.23 in health, meaning that “one fifth of the entire extracullar volume is to be found in the plasma”. Notice, however, that the concept of PV:IF ratio is emphasised in Moore’s commentary in order to highlight the fact that it is not a fixed ratio, it changes during hemorrhage, during spontaneous plasma refill, and during the infusion of intravenous fluids .

Writing in 1985, Twigley and Hillman expressed the view that contemporary perioperative fluid prescription was still predominantly driven by Shires’ third space hypothesis and an understandable fear of renal failure and hyperkalemia [7]. They proposed the startlingly simple solution that blood and colloid solutions should be more used in order to preserve the blood volume and renal perfusion while restricting crystalloid infusion and so avoiding edema. Their “new approach to perioperative fluid administration ” has been much repeated, with minor variations, in journals and textbooks by many of today’s fluid therapy experts. It views the total body water as existing in three static compartments: (1) the plasma (with a suspension of red cells), (2) the interstitial space, and (3) the intercellular space. Infused colloid solutions are said to be restricted to the plasma by largely impermeable capillary walls, isotonic salt solutions are said to distribute through the plasma and interstitial compartments as far as the cell membrane, while free water dilutes all three compartments [8]. As we shall see, this paradigm is too simplistic. It does not explain what is observed in clinical practice and so cannot be used as the basis of a rational prescribing strategy. On the electrolyte front it ignores the fact that the volume of distribution of sodium is close to the total body water (the Sodium Paradox), making an isotonic salt solution an appropriate choice to treat pure dehydration [9]. On the colloid osmotic pressure front, it is incompatible with the Starling principle. Rodney Levick and Charles Michel explained some clinical corollaries of a revised Starling principle in 2010 [10]. Michel now prefers the term steady-state Starling principle . It was confirmed by experiments published in 2004, but had largely been ignored by clinicians and physiology teachers, though it held the key to explaining much clinical and experimental data that were otherwise inexplicable [11, 12]. The widely taught model of filtration–reabsorption balance does not occur in the microcirculation of most tissues. Tissue fluid balance depends on lymphatic function in most tissues [10].

My son and I offered a better paradigm, based on modern cardiovascular physiology. We called it the revised Starling equation and glycocalyx model (RSE&GM) [13]. It is a working paradigm for clinicians that explains the previously inexplicable: Why is 100 ml of isotonic salt solution as effective for resuscitation as 62–76 ml human albumin solution or 63–69 ml of a colloid solution [14, 15]? It explains why albumin bolus therapy is no more effective than isotonic salt solution in shock therapy, and why fluid boluses may be harmful [16, 17]. It explains why extravascular lung water may increase during fluid loading in the critically ill with presumed hypovolemia [18, 19].

An expert group writing in 2014 opined that “improved understanding of the endothelial glycocalyx has altered our comprehension of the role of colloid osmotic pressure in fluid balance” [20]. Since the first edition of this textbook the clinical relevance of the steady-state Starling principle has gained wider acceptance [21]. Nonetheless, more liberal approaches to fluid therapy, and a teaching that fluid overload and tissue edema are somehow inevitable during the early phase of critical illness and in the perioperative phase of major surgery may still be harming patients. In the second edition of this chapter, I update the evidence supporting RSE&GM and examine other recent developments in fluid physiology.

The Michel–Weinbaum Model

The glycocalyx model, also known as the Michel–Weinbaum glycocalyx junction-break model of fluid exchange in a continuous capillary, is key to the new physiology. It is described by one of the men for whom it is named in Chap. 2. The microvascular permeability barrier of most capillary beds is revealed to function like a one-way valve that permits filtration but does not permit reabsorption (Fig. 9.1). At steady state filtration the counter-current diffusion of interstitial proteins is too slow to raise the subglycocalyx colloid osmotic pressure π_g which is therefore lower than the general interstitial colloid osmotic pressure π_i, creating a substantial plasma to subglycocalyx colloid osmotic pressure difference opposing the filtration rate. When capillary pressure falls low enough for reabsorption to begin, diffusion of interstitial proteins to the subglycocalyx space wipes out the colloid osmotic pressure difference that was driving reabsorption and a new steady state of minimal filtration is established. Notice that the endothelial surface layer that excludes red blood cells has two phases, with intravascular albumin molecules concentrated at the interface between the membrane-bound glycocalyx matrix and the outer surface layer.

Fig. 9.1

A cartoon of solvent and protein movement through an interendothelial cleft during filtration (above) and transient reabsorption (below). The concentration of proteins (shown as dots) is very low in the subglycocalyx space during filtration and contributes to a very high colloid osmotic pressure difference across the glycocalyx which normally opposes filtration

The Revised Starling Equation and Glycocalyx Model (RSE&GM) Paradigm

The revised Starling equation and Glycocalyx model (RSE&GM) paradigm is founded on clinical experiments by Robert Hahn who showed that the rapid infusion of an isotonic salt solution brings about a change in red blood cell dilution that can be explained by a two-compartment volume kinetic model [22,23,24]. There is an initial central volume of distribution for the infused fluid and a subsequent distribution to a peripheral or tissue volume. The infusion has measurable effects on transendothelial filtration, lymph flow and urine output which appear to depend on the rate of infusion [24] (Fig. 9.2). Hahn reports that colloid solution boluses have a more profound and prolonged effect on red cell dilution, making a single compartment model adequate for describing the kinetics of the acute volume disequilibrium. That is not to say, however, that colloid therapy has no effect on transendothelial filtration and lymph flow.

Fig. 9.2

A kinetic diagram of extracellular fluid distribution. Gel phase of the central and tissue volumes indicated by yellow, free-flowing plasma of the central volume indicated by pink. Erythrocyte volume indicated by red disks. (a) Shows a healthy intravascular gel phase (the endothelial glycocalyx and outer surface layer). (b) Notice how dehydration or fragmentation of the endothelial glycocalyx layer leads to erythrocyte dilution. When induced by colloid infusion, this dilutional anemia is frequently but erroneously stated to prove that colloids are superior plasma volume expanders

RSE & GM emphasizes volume kinetics. RSE&GM brings several new physiological concepts to the fluid debate, which help to explain why colloids are of little benefit for resuscitation from hypovolemia. It is important to remember that I am here presenting a clinician’s simplified approximation of the relevant physiology. I provide a comprehensive account of modern fluid physiology for clinicians in a recent textbook [25].

Heterogeneity of the Microvasculature

Fluid therapy teaching often overlooks the heterogeneity of capillaries and venules. The human microvasculature is composed of four very different types of capillaries, whose distinct functions must be appreciated:

• 1. The 1.5 kg liver takes around 1 ml/min of the healthy cardiac output/gram of tissue. Like the spleen and bone marrow, it has “sinusoidal” capillaries , which have an incomplete glycocalyx and are freely permeable to larger molecules, making the interstitial fluid of these tissues an extension of the plasma volume. It creates about 50% of the total lymphatic flow to the thoracic duct. In resuscitated septic shock patients, as much as 50% of the cardiac output goes to this very leaky microcirculation.

• 2. The highly specialized renal glomerular capillaries are continuous but feature open fenestrations for the filtration of fluid to the renal tubules (the glomerular filtration rate [GFR]). They operate at a high capillary pressure and never absorb fluid.

• 3. Diaphragm-fenestrated capillaries operate at lower capillary pressures and are specialized to absorb fluid from interstitium to plasma when needed. They are found in absorbing tissues such as the intestinal mucosa, the endocrine glands, lymph nodes, and renal peritubular capillaries.

• 4. The greatest number of systemic and pulmonary capillaries are nonsinusoidal, nonfenestrated capillaries and feature a continuous endothelial surface layer that sparingly filters fluid via occasional gaps in interendothelial cell junctions to the interstitium of their tissues, including connective tissues, lung parenchyma, and brain. The endothelial surface layer is now appreciated to have two phases. The membrane-bound glycocalyx molecules support a more fluid outer zone which features long strands of hyaluronan. The Michel–Weinbaum glycocalyx junction-break model proposes that effective pore size (small or large) in such capillaries is a function of the spaces between the matrix fibers of the glycocalyx, while the area for fluid exchange is a function of the length of the slit-like junction breaks between adjacent endothelial cells. In the capillaries of the brain and spinal cord, endothelial cell membranes are tightly opposed by zona occludens tight junctions with few breaks, resulting in very small effective pore size of barely 1 nm. Their integrity is reinforced by the presence of adjacent interstitial pericytes (microglia). The blood–brain barrier is therefore only permeable to the smallest non-lipid-soluble molecules. Nonsinusoidal, nonfenestrated capillaries of muscles, connective tissues, and lungs have macula occludens loose junctions to their intercellular clefts, and the effective pore size there is up to 5 nm, making them permeable to molecules as large as myoglobin.

Now consider the endothelial surface layer in these capillaries. It is easily compressed in volume by almost anything we do to our patients, releasing glycosaminoglycans (GAGs) into the circulating plasma. Albumin molecules are entrapped and retained at the interface between the glycocalyx and its outer layer, in part by electrostatic means, restricting their free passage through it, and so, behaving as an imperfect filter. Inflammation damages the glycocalyx and, so, reduces the reflection coefficient sigma for albumin, with glycocalyx components including the GAGs appearing in blood samples. Staverman’s reflection coefficient sigma is an index of the effective, as opposed to measured, colloid osmotic pressure effect on transendothelial solvent filtration (J_v) . As such, a reduced sigma limits the colloid osmotic pressure opposition to J_v and gives free rein to the hydrostatic pressure filtration driver. However, even when sigma approaches zero, the resistance to fluid filtration caused by the basal membrane and extracellular matrix gel remains substantial. The tissues that can accumulate substantial amounts of interstitial fluid after trauma and sepsis (i.e, the more compliant tissues) are loose connective tissues, muscles, lungs, and gastrointestinal mesentery and mucosa. For example, extravascular lung water measured by double-indicator dilution can increase from around 500 ml to 2.5 l in pulmonary edema, while the loose connective tissues and muscles can expand to many liters of peripheral edema.

Hahn’s Volume Kinetics

Robert Hahn’s series of experiments on the volume kinetics of rapidly infused intravenous (IV) fluids measured the hemoglobin concentration of arterial or venous blood and modeled a central and a tissue volume that broadly represents the intravascular and extravascular fluid volumes, respectively. The peripheral volume is found to be 6–8 l, less than the anatomic interstitial fluid volume. As volume kinetics measure only the volume that can be expanded, this will therefore not include spaces limited by rigid structures such as the bone (brain, marrow) or fibrous capsules (liver, spleen, kidney) (Fig. 9.1). This goes some way to explaining why isotonic salt solutions are more efficient plasma expanders than we might expect if we were to presume their distribution throughout the total body water. In systemic capillary leak syndrome, so much fluid goes to the soft tissues of the limbs that it can cause compartment syndromes.

Three Intravascular Fluid Volumes

For the purposes of the RSE&GM paradigm, there are three intravascular fluid volumes. The first two, plasma volume and red cell volume, which make up the circulating blood volume, are well known. The third is the noncirculating intravascular volume occupied by the endothelial surface layer which includes the membrane-bound glycocalyx molecules that create a fiber matrix scaffold). Being on average about two microns thick, the fragile endothelial surface layer can account for as much as 1.5 l of the intravascular volume in health. As it excludes red cells, acute reductions in the thickness (and so volume) will increase the volume available to red cells and lead to a reduction in the hematocrit. GAGs are typically shed to the circulation when the glycocalyx is thinned.

A key concept of the new paradigm is that a bolus of an isosmotic plasma substitute has a central volume of distribution that approximates the free-flowing plasma, while a bolus of an isotonic salt solution has a central volume of distribution that includes the intravascular gel phase and approximates the whole of the intravascular volume, and possibly the interstitial space of the sinusoidal tissues. The concept is supported by consistent clinical reports that adequate resuscitation with an isosmotic plasma substitute can be achieved with slightly smaller volumes than adequate resuscitation with a crystalloid, but at the expense of much diluted hematocrit. The ability of plasma and plasma substitutes to cause anemia is still widely misinterpreted as indicating that the colloids are “better volume expanders.” In volunteers and anesthetized patients, the volume effect ratio crystalloid to colloid for causing anemia and hypoalbuminemia is about 4:1 or 5:1. In the misleading words of researchers in Munich, “The intravascular volume effect of Ringer’s lactate is below 20%” [26]. However, in septic patients, in nonseptic intensive care patients, and in surgical patients undergoing goal-directed fluid therapy, the volume effect ratio crystalloid to colloid for correction of hypovolemia is only about 1.5:1. The data of Jacob et al. [26] show that a crystalloid infusion at three times the rate of blood withdrawal perfectly preserves hemodynamic stability; in other words, they could have concluded that for hemodynamic purposes “the intravascular volume effect of Ringer’s lactate is greater than 33%”. Erythrocyte or albumin dilution data were wrongly presumed to indicate resuscitative effectiveness.

Biophysical Osmotherapy Causes Haemodilution

It has long been known that plasma volume refill after haemorrhage or withdrawal of blood occurs at a rate of about 1 ml per minute and is achieved without a reduction in plasma protein concentration. Most of this autoresuscitation fluid comes from pumped mesenteric lymphatics and the thoracic duct, and is called “the essentiality of the lymphatic system to the recovery from shock” [27,28,29,30].

To investigate if edema fluid can be mobilised from the tissues to the blood and excreted as urine, researchers at the Karolinska Institute recently investigated the hemodilution that follows an acute increase in the plasma colloid osmotic pressure in fifteen healthy euvolaemic volunteers. Using a hyperosmotic human albumin solution they increased the plasma colloid osmotic pressure by about 8%, the plasma albumin concentration by about 16% and found that the volume of distribution of red blood cells (their surrogate for plasma volume) was transiently increased by about 16%. Maximal haemodilution was reached within 20 min post-infusion but was then halved by 4 h [31]. Red cell dilution studies of hyperoncotic human albumin solution transfusion in critically-ill patients have also been interpreted as showing osmotic absorption of fluid from the extravascular to intravascular compartment, but without information from an indicator of the whole intravascular volume, such as Dextran 40, such a conclusion is not justified. To cause an acute increase in circulating plasma colloid osmotic pressure is to cause a disequilibrium event that first draws water from the non-circulating gel phase of the intravascular volume associated with the glycocalyx and the interstitial fluid of the liver. The steady-state that follows is of greater importance but is too often overlooked.

After a century of the classic Starling principle, which suggests filtration at the arteriolar portion of a capillary and reabsorption at the venular end, it can be hard to accept the slightly more complex reality and to grasp the consequences that inform RSE&GM. With the exception of the diaphragm-fenestrated capillaries that can absorb solutes at normal capillary pressure, reabsorption of fluid from interstitium to plasma does not occur, even at reduced capillary pressure. The mechanism is the Michel–Weinbaum glycocalyx junction-break model, which preserves a state of minimal filtration even when the hydrostatic transendothelial pressure difference delta P is low. Intravenous colloid therapy cannot promote absorption and cannot help to prevent or treat interstitial edema.

The J-Curve and the J-Point

As a consequence of the Michel–Weinbaum glycocalyx junction-break model, a plot of the transendothelial solvent filtration rate J_v against capillary pressure based on the steady-state Starling principle demonstrates that J_v remains close to zero with rising capillary pressure until the convection current of filtrate through the interendothelial channels is sufficient to bring the sub-glycocalyx-protected region’s colloid osmotic pressure π_g close to zero. The transendothelial colloid osmotic pressure difference delta π is then maximal, and further increases in delta P will widen the difference between delta P and the now-fixed delta π, causing a sharp rise in J_v. This creates an inflection on the curve that makes it appear J-shaped (Fig. 9.3). The inflection is called the J-point [13]. Charles Michel in this textbook describes this curve as a hockey stick curve.

Fig. 9.3

The J-curve. A plot of the transendothelial solvent filtration rate J_v against the capillary pressure P_cap. In health, capillary pressure is regulated around the J-point. Raising the colloid osmotic pressure of plasma π_p only increases the transendothelial colloid osmotic pressure difference delta π when capillary pressure is higher than normal (heart failure or fluid overload) and the sub-glycocalyx colloid osmotic pressure π_g is close to zero. Around and below the J-point (health and hypovolemia), J_v is low and essentially uninfluenced by π_p

Manipulating Capillary Pressure

One of the first consequences of inflammation is a fall in the interstitial pressure as integrins change the conformation and hydration of structural collagen fibers. J_v therefore increases, beginning the shift of extracellular fluid balance from the intravascular to the extravascular compartment. Precapillary vasodilation follows and increases the capillary pressure and so further increases J_v. If the hemodynamic reflexes are working, blood pressure is maintained by increased cardiac output (capillary recruitment), which is another factor increasing J_v in the early phase of systemic inflammation.

A focus on capillary pressure brings a new perspective on the role of low-dose arteriolar pressor therapy in anesthesia and intensive care practice. Typically, the goal of pressor therapy is to maintain an adequate systolic arterial pressure after adequate resuscitation of the intravascular volume and stroke volume, allowing more restrictive use of fluids. In terms of the RSE&GM, arteriolar pressors are expected to raise diastolic arterial pressure but lower capillary pressure and so reduce J_v, keeping more of the extracellular volume intravascular. Now we can see alpha-1 agonists as part of a potential anti-edema strategy. They have even been found to increase urine output, but I do not think it is helpful to describe this as a diuretic effect. Notice that in vasodilated contexts, such as sepsis or anesthesia [32], capillary filling pressure is close to the arterial diastolic pressure. There is a case for paying more attention to diastolic pressure in perioperative care [33, 34].

Concern that capillary hypertension is injurious will make us more cautious about employing rapid boluses as a way to increase diastolic pressure. RSE&GM predicts transiently high capillary pressures during rapid transfusion, which will cause excessively raised J_v, reducing the intravascular contribution of whatever resuscitation fluid we choose. A slower infusion rate will cause lower capillary pressure peaks, minimizing hyperfiltration and maximizing efficient resuscitation of the intravascular volume. Robert Hahn has produced human data that confirm the prediction, but he attributes the phenomenon to the visco-elastic properties of the interstitial matrix . If we infuse fluids at lower rates, edema is avoided and the central fluid volume is better conserved.

Focusing on capillary pressure through the RSE&GM paradigm reveals a need to understand the transendothelial resistance to fluid flux (represented in the Starling equation as its reciprocal, the hydraulic conductance L_p). The glycocalyx is the first and the major fiber matrix resistor in the current of fluid and solutes between plasma and lymph. The basement membrane and extracellular matrix are the second and third resistances in a series.

The basement membrane, where it exists, is a specialized part of the extracellular matrix 60–100 nm in thickness, composed of type IV collagen and laminin and closely adherent to the cell membrane. The collagen matrix can be thought of as a special phase of the extracellular fluid, which provides an exchangeable sodium store (around 400 mmol/l, compared to about 145 mmol/in general interstitial fluid) [35, 36]. Bhave and Neilson propose that short-term sodium storage and interstitial volume homeostasis may be relevant to transient or disequilibrium phenomena such as blood pressure dipping, flash pulmonary edema, rapid blood loss, burns, and sepsis [37]. It may be a mechanism that enables hypertonic sodium resuscitation to be effective without edema or hypernatremia.

Collagen fibrils also occur within the interstitial space, upon which glycoproteins such as fibronectin and proteoglycans (protein molecules with GAG side chains) are arranged, and contain free GAGs. Toll-like receptors are found within the extracellular matrix and are believed to have a pivotal role in the early development of systemic inflammatory response and ventilator-induced lung injury. Integrins and their receptors modulate cell locomotion through the extracellular matrix, and it has been discovered that they can modulate interstitial pressure by bringing about conformational changes to collagen that allow the GAGs to become hydrated. An acute reduction in interstitial pressure occurs in inflammatory conditions, increasing delta-P and thereby increasing J_v by as much as 20-fold independently of other causes of capillary “leak.” Changes that compact the glycocalyx releasing GAGs into the circulating plasma are associated with increased transendothelial protein flux, but compaction of the glycocalyx and increased porosity may be separate processes and the association may not be entirely causal. Although transfused macromolecules do not easily permeate an intact endothelial glycocalyx layer, they pass easily into the interstitial fluid of the sinusoidal capillaries in the bone marrow, spleen, and liver, equilibrating with interstitial macromolecules and returning to the venous system via lymphatics. An increase in the proportion of the cardiac output going to sinusoidal tissues will increase overall J_v and the transcapillary escape rate of albumin.

Understanding “Leaky Capillaries”

Many of the clinical challenges of anesthesia and critical care are attributed to what we often call “leaky capillaries.” Pushed to elaborate, even the experts may recall the reflection coefficient but few can elaborate further. The equations we are told are Starling’s were in fact proposed many years after his death. The most frequently cited equation explains the transendothelial solvent filtration rate J_v in terms that describe the net hydrostatic pressure difference and net colloid osmotic pressure difference across the semipermeable microcirculation. The reflection coefficient sigma modifies the apparent delta π to the effective delta π. Histamine and other autocoids are known to increase the length and number of intercellular junction breaks, especially in the distal part of the capillary and the venules. The surface area for filtration within each capillary and the hydraulic conductivity L_p are thereby increased and increase the J_v. Note that in some versions of the Starling equation for J_v, the product of surface area and hydraulic conductivity is called the filtration coefficient K_fc.

Less often taught, but equally important to understanding the pathophysiology, is the equation explaining a transendothelial solute transfer rate J_s as the sum of the mass of that solute carried with the transendothelial filtrate (convection) and the mass of that solute that permeates the microcirculation independently of flow (diffusion). In clinical considerations the solute of interest is albumin. Researchers who measure J_s of albumin or another marker molecule in disease states often presume that J_v will be increased with J_s and cause edema. Permeability as it appears in the equation for J_s is an index of how readily albumin appears to diffuse across a capillary if it were a simple semipermeable membrane dividing static fluid spaces. It is not. Albumin is actively transported across continuous capillaries via a membrane-associated protein that has been called gp60 or PV-1 and is now referred to as caveolin. Caveolin deficiency is incompatible with life. The rate of transfer of albumin, and other proteins having their own transport system, to the interstitium will appear to be a change in the number of large pores. Plasmalemmal vesicles (caveolae) carry some water with the albumin, but the convective interendothelial pathways predominate. Places where the fibrematrix covering a junction break is thinned will also behave like more large pores.

Curry and Adamson have reviewed understanding of the tonic regulation of vascular permeability in health and disease [38]. Sphingosine-1-phosphate is synthesized in erythrocytes and transported to the endothelium by albumin. It modulates

• the adherens junction,

• continuity of tight junction strands,

• and the synthesis and degradation of glycocalyx components.

Baseline permeability appears to be maintained by the small GTPase enzymes called Rap1 and Rac1, which are dependent upon the supply of sphyngosine-1-phosphate. Inflammatory stimuli act to reduce Rac1 and Rap1 activity, and so enhanced delivery of sphyngosine-1-phosphate should be able to buffer inflammatory harm. This knowledge suggests it is important to maintain erythrocyte and endogenous albumin delivery to the vascular endothelium and should make us even more concerned about the use of plasma substitutes that cause anemia and hypoalbuminemia.

The Circulation of Tissue Fluid to Lymphatic Vessels and Return to the Intravascular Space

RSE&GM recognizes that the microvasculature is not a passive biophysical barrier separating the vascular and interstitial compartments of the extracellular fluid’s circulation. The collecting (afferent) lymph vessels have barrier properties comparable to the venules and carry filtered tissue fluid to the lymph nodes whose capillaries are diaphragm-fenestrated and capable of fluid absorption. As much as 50% of the fluid arriving at a lymph node is reabsorbed there, so the lymph in the efferent lymphatics has a high protein concentration and is pumped to the thoracic duct. It is thought that most of the efferent lymph re-enters the venous system via the thoracic duct, but other lymphatic–venous collaterals can be recruited if the duct is tied off. Radiation ablation of lymph nodes predisposes to edema, a clear practical demonstration that nonfenestrated capillaries outside the lymph nodes are not capable of significant absorption of tissue fluid. The spontaneous contractility of the lymphatics is enhanced by alpha adrenergic agents and suppressed by inflammatory mediators and opioids. It is worth recalling that 25% of the cardiac output goes to the discontinuous capillary circulations of the liver, spleen, and bone marrow, where sigma for albumin and any other large molecule is very low, and that more than 50% of the high-protein lymph in the thoracic duct originates from the liver. In resuscitated hyperdynamic sepsis, the proportion of blood going to the liver rises to as much as 50% so that higher-molecular-weight molecules will be easily lost from the bloodstream.

Missing Sodium

It is widely appreciated that renal sodium reabsorption under the influence of aldosterone occurs in the distal convoluted tubules and maintains both extracellular fluid volume and osmolarity. Excess sodium intake drives up extracellular fluid volume with adverse consequences. Fluid and electrolyte balance studies on perioperative patients reveal a rather more complex state of affairs. The volume of distribution of sodium within the body is much larger than the extracellular fluid volume. Surgical patients often acquire a substantial sodium load without the associated water retention (edema) we would expect from the osmolar balance—this has been called “missing sodium”. This non-osmolar sodium appears to be stored within the gel phase of the interstitium, and especially the interstitium of skin and connective tissues, as sodium hyaluronan. Non-osmolar sodium is therefore in an equilibrium with osmolar sodium ions and is being shown to have a physiological role in fine-tuning the extracellular fluid volume while sodium balance fluctuates. Moreover, longer term sodium loading leads to increased density of interstitial lymphatic channels which are the active interface between the stored and the circulating sodium. It has often been taught that excess sodium intfusion inevitably leads to edema, but the non-ionic non-edematous storage of excess sodium in surgical patients was first observed in 1986 [39]. The clinical importance of this second sodium store is emerging.

Intracellular Fluid Volume Is Regulated Independently of Total Body Water

A parallel phenomenon is the maintenance of intracellular fluid volume over a range of body water osmolarities. In a cohort study of post-surgical patients treated in an Intensive Care Unit with conventional intravenous fluids for 4 days, Hessels and colleagues found that there was a strongly positive accumulation of sodium and total fluid, but a negative balance of electrolyte-free water and potassium. In a sub-study comparing the effects of a prescribing potassium to a target of [4.0 mmol l⁻¹] or [4.5 mmol l⁻¹] they found that all the excess potassium of the second group was renally excreted. They reasonably interpreted these observations as showing that excess fluid in clinical practice results in interstitial expansion (extracellular edema) while the intracellular volume, where potassium is the dominant osmolar cation, is regulated close to its healthy normal. They speculate that the cytosol is able to clear alternative osmolytes when there is volume increase by electrolyte free water infusion, and generate alternative osmolytes when hypertonic saline infusion reduces cell volume. Intracellular volume is thereby conserved in the face of changing body water tonicity [40, 41].

A Revised Twigley–Hillman Diagram

The Twigley–Hillman diagram represents the total body water in three static compartments divided by two barriers [7, 8]. I have offered an improved 2-compartment volume kinetic representation of extracellular fluid to assist rational fluid prescription. The volume of the third compartment, intracellular fluid, remains remarkably constant with fluctuations in the extracellular fluid volume [40, 41]. I suggest a revised Twigley–Hillman diagram of the total body water. From left to right, it represents:

• Plasma. The free-flowing intravascular plasma volume. Normally around 3 l of the 4 l intravascular fluid.

• Red blood cell volume (intracellular fluid) of approximately 2 l is not shown separately on this diagram.

• Endothelial surface layer (ESL). Circulating erythrocytes are repelled by the outermost, rather porous gel phase of the endothelial surface layer while the inner fibre matrix of the membrane-bound glycocalyx performs the small pore function of excluding larger molecules, including albumin. It is impossible to attribute a precise volume to this phase, but about 1 l in health.

• Aqueous and gel fluid phases of the triphasic interstitium are distinguished. In health the aqueous phase may be a little as 1%, but increases greatly in oedema. The third phase is structural, mostly collagen Type I fibres, which appear to channel the interstitial flow of solvent and solutes. Sodium hyaluronan is a non-osmotic buffer of sodium ions in the gel phase, and albumin is excluded.

• Lymph is aqueous interstitial fluid that has entered the lymphatic vasculature.

• Intracellular fluid.

The revised Twigley–Hillman diagram (Fig. 9.4) of total body water compartmentalisation emphasises the extracellular fluid (ECF) circulation (blue arrows) which occurs in most tissues most of the time. The intravascular space normally contains about 5 litres of blood and 1 litre endothelial surface layer from which circulating red blood cells are excluded. The intravascular extracellular fluid is free-flowing aqueous (plasma) and gel phase (ESL). ESL contains the fibre matrix molecules of glycocalyx which arise from the endothelial cell surface. The triphasic interstitial space has a structural collagen fibrous phase and around 14 litres of fluid; an aqueous phase, a gel phase within which glycosaminoglycans such as hyaluronic acid have the capacity to store sodium without raising tissue osmolality, and lymph. The intracellular fluid (ICF) volume, normally about 23 litres, is sensitive to acute changes in ECF osmolality. However, cell volume regulatory mechanisms exist to preserve the steady-state intracellular fluid volume and enable subjects to tolerate chronic hypotonicity.

Fig. 9.4

A revised Twigley–Hillman diagram. J_v is the transendothelial solvent filtration rate which at steady state is balanced by the lymphatic return of solvent Q_lymph via capillary absorption within lymph nodes and as efferent lymph

Water and solutes enter and leave the body across epithelial barriers, represented here by a brown ellipsoid. In clinical practice we can infuse fluids directly into, or haemofilter water out of the free-flowing plasma.

In Sinusoidal tissues (liver, spleen and bone marrow) the ESL is discontinuous and there are windows (fenestrations) through the endothelium that exclude red blood cells but admit albumin to the interstitial space so that there is no trans-endothelial colloid osmotic pressure difference to oppose filtration.

In Non-sinusoidal tissues the continuous ESL is almost impermeable to albumin so that filtered fluid in the immediate sub-glycocalyx space (protected region) has a very low colloid osmotic pressure compared to plasma or the general interstitial fluid. The trans-endothelial colloid osmotic pressure difference opposing filtration is therefore high. Michel and Weinbaum correctly hypothesised that if transendothelial water movement across non-sinusoidal capillaries and venules is transiently reversed by a sudden drop in the hydrostatic pressure difference, interstitial albumin rapidly enters the protected region, diminishing the trans-endothelial colloid osmotic pressure difference. The net water movement therefore quickly return to steady-state filtration. This is the Michel–Weinbaum No Reabsorption Rule.

Water is absorbed from afferent lymph into lymph node capillaries and venules. Efferent lymph therefore has high protein and lipid content and returns to the central veins via the thoracic duct. Acceleration of protein-rich efferent lymph to the circulating blood volume is an important compensatory response to haemorrhagic shock in humans.

Exceptions to the No Steady State Absorption Rule: Hypodermoclysis

In 1896 Ernest Starling performed experiments on small dogs (typically 6–12 kg in weight) in which he infused 1% sodium chloride solution into the connective tissues of the back leg and observed that this fluid was absorbed into the blood stream by capillaries [42]. He wrote;

The importance of these measurements lies in the fact that, although the osmotic pressure of the proteids of the plasma is so insignificant, it is of an order of magnitude comparable to that of the capillary pressures; and whereas capillary pressure determines transudation, the osmotic pressure of the proteids of the serum determines absorption.

This science is still emphasised in undergraduate medical education and in postgraduate critical care teaching, but Starling himself added some caveats. Firstly, he appreciated that the effect was transient and would lead to a no-absorption steady state;

With diminished capillary pressure there will be an osmotic absorption of salt solution from the extravascular fluid, until this becomes richer in proteids; and the difference between its (proteid) osmotic pressure and that of the intravascular plasma is equal to the diminished capillary pressure.

Secondly, he appreciated that the effect he observed with exogenous saline may not be applicable to what he called dropsical fluid reabsorption;

Salt solutions, isotonic with the blood-plasma, can be and are absorbed directly by the blood vesels. This statement probably holds good for dropsical fluids containing small percentages of proteids.

A century on the only fault that we can find in Starling’s interpretation of his experiments is that he did not entirely appreciate the importance of the fact that the interstitial proteid concentration is dynamically determined by the rate of solute influx, J_s, relative to the rate of water influx J_ν. The design of his experiment uncoupled the physiological inverse ‘extravascular dilution’ relationship between or π_i and J_ν, and created an exceptional condition that has misled colloid osmotic pressure therapists ever since.

To this day the subcutaneous route is known to be safe and effective for the administration of isotonic solutions and certain drugs. Indeed, the complication rate is less than with intravenous administration. Hypodermoclysis is widely used in smaller animal veterinary practice, but in an adult human two litres per day is achievable, enough for what is called ‘maintenance’ therapy while enteral nutrition and hydration are suspended. The rate of absorption can be increased by adding hyaluronidase, an enzyme that increases the area for absorption of fluids and drugs. Hypodermoclysis with intravenous fluids brings about the same electrolyte measurements or osmolalities as intravenous infusion at the same rate. Animal studies have shown equivalent subcutaneous absorption of cephalosporins and clindamycin when compared with intravenous absorption. Studies with radioisotope tracers in humans have shown that clearance from subcutaneous tissue is complete within an hour after the termination of the subcutaneous infusion [43].

Recall that fluid filtered by the glycocalyx layer is almost protein-free and creates a low colloid osmotic pressure “microdomain” at the interendothelial cleft exit. However, if filtration slows, tissue proteins almost immediately diffuse back into the cleft, raising the colloid osmotic pressure, diminishing the colloid osmotic pressure difference, and preserving a low rate of filtration. In the case of injected protein-free isotonic salt solution, the volume of the low colloid osmotic pressure domain at adjacent cleft exits is relatively immense, and there is no available protein to diffuse back into the cleft. Absorption can therefore occur until the injected volume has been taken up and tissue proteins can again enter the cleft, restoring the normal equilibrium of low filtration.

Exceptions to the ‘No Steady-State Absorption’ Rule: Effect of Local Epithelial Transport

There are of course situations where absorption of fluid from the interstitium to the plasma is physiologically vital.

• The peritubular capillaries of the renal cortex and the ascending vasa recta of the renal medulla are in a continuous state of absorption of interstitial fluid which is continuously secreted by renal tubular epithelium.

• The renal collecting ducts supply protein-free solvent and solutes to the medullary interstitium to keep the colloid osmotic pressure low enough for continuous absorption of fluid by the medullary capillaries. Convective solvent flow even carries interstitial albumin molecules back into the renal veins. There are no lymphatic channels in the renal medulla.

• While the intestine is absorbing enteral water the mucosal epithelium secretes water into the interstitium for absorption by mucosal capillaries.

• In lymph nodes interstitial fluid is continuously replenished by the flow of pre-nodal lymph with a low protein concentration.

• At the blood-brain barrier the perivascular pool of aquaporin 4 channels in pericytes (astroglial cells) delivers water to the abluminal side of the endothelium. At the same time cerebrospinal fluid is being delivered by the glymphatic system to the aquaporin-4 channels of astroglial cell foot processes so that it can join the brain interstitial fluid.

In each case the renewal of interstitial fluid from an adjacent epithelial secretion of low protein concentration uncouples the inverse ‘extravascular dilution’ relationship between π_i and J_ν, and prevents π_i from approaching π_p. The colloid osmotic pressure difference supporting absorption of interstitial fluid to the plasma is protected.

Exceptions to the ‘No Steady-State Absorption’ Rule: Discontinuous Capillaries of the Sinusoidal Tissues

There is almost no hydrostatic pressure gradient or colloid osmotic pressure gradient being exerted across the fenestrated endothelial cells of the space of Disse, the hepatic version of interstitial space. It is freely connected to the intravascular space and so solvent and solutes may freely pass in either direction. Indeed, if the purpose of a lymphatic system is to circulate interstitial proteins and larger molecules from intestitium to plasma, one might question whether the liver needs a lymphatic system at all. It has long been appreciated that fluid infused into bone marrow equilibrates very rapidly with the plasma volume, without having to go via lymphatics.

Scientific Method in Peri-operative Fluid Therapy Research

In the first edition of this Chapter I pointed out that the great philosopher Ludwig Wittgenstein was very critical of the Scientific Method as applied to investigation of the condition called shock back in 1943 [44, 45]. The scientific method starts with the observation of how things are. Rational fluid therapy requires us to understand the contribution of suboptimal fluid therapy to surgical complications. Do complications caused by suboptimal fluid therapy (excluding negligent failure to treat or willful overdose) lead directly to surgical mortality? Ghaferi and colleagues at the Centre for Health Care Outcomes and Policy observe in various surgical patient populations that complication rates are broadly similar between hospitals with very low mortality rates and very high mortality rates [46]. From the information they gather, they move to the second step of the scientific method and propose a hypothesis. It is their hypothesis that higher surgical mortality rates across hospitals are largely explicable by an institutional failure to rescue patients who develop complications. The third step of the scientific method is to test that hypothesis in a reproducible controlled experiment, but there are clear ethical difficulties here. Fluid therapy researchers have to focus on complication rates rather than mortality, as they are far more common and are presumed, rightly or wrongly, to lead to mortality, as well as increasing the costs of achieving the desired final outcome. They have designed randomized controlled trials (RCTs) of protocol A versus protocol B but they are rarely reproducible. The fourth step, analysis of the data from the experiment, usually fails to deliver incontrovertible conclusions, and the underlying hypothesis is rarely questioned. The ultimate step will be to reproduce experiments that deliver confirmation of hypotheses.

A classification of post-surgical complications, first published in 1992, was critically re-evaluated, modified and retested to increase its accuracy and its acceptability in the surgical community. The new Clavien–Dindo Classification was prospectively validated in a cohort of 6336 patients who underwent elective general surgery [47] and has gone on to gain wide acceptance as a system for recording complications and their severity [48]. A recent clinical trial of a nutritional intervention in patients undergoing elective colorectal surgery with primary anastomosis at six clinical centres in the Netherlands and Denmark found the commonest adverse events to be postoperative ileus (20–30%), anastomotic leakage (circa 10%) and pneumonia (less than 10%) [49]. Fluid overload would be suspected as a possible contributor to such complications. Intraoperative Continuous Intestinal Loop Warming (ICLW) reduces the incidence and duration of post-surgical ileus [50].

Varadhan and Lobo made a very helpful contribution to definitions in their meta-analysis of nine randomized controlled trials of intravenous fluid therapy in major elective open abdominal surgery [51]. They defined restricted fluid therapy as less than 1.75 l of “maintenance fluid” per day. Liberal fluid therapy was defined as more than 2.75 l per day for “maintenance fluid” per day. As Twigley and Hillman had commented 25 years earlier, the “standard” maintenance fluid prescription is based on the needs of a physiologically unstressed adult and is typically 3 l of water and 150 mmol sodium per day, which falls within their definition of liberal. Finding no difference in the complication rate between liberal and restricted strategies, Varadhan and Lobo reclassified patients according to fluid balance; balanced being zero volume balance with little or no weight change. Patients exposed to imbalance appeared to experience a higher complication rate. Their hypothesis is supported by bio-impedance data confirming the expectation that fluid imbalance is associated with ascites or fluid collections during the postoperative period in patients undergoing hepato-pancreatico-biliary operations [52]. Zero-balance, or euvolemic goal-directed fluid therapy (GDFT) , is increasingly seen as good perioperative practice [53].

Clinical Research

Doherty and Buggy reviewed the evidence on perioperative fluid therapy volumes up to 2012 [54]. There was very little modern research with which to work. In 2003, Brandstrup’s Danish multicenter study randomized 172 patients undergoing colorectal surgery to restricted (zero-balance) or liberal (anesthetist’s standard) fluid therapy [55]. There were fewer complications and no deaths in the restricted group, while the standard group included four deaths and more complications. Mackay had found no advantage for a restricted protocol in a trial that randomized 80 patients undergoing colorectal surgery [56]. In Denmark, Holte randomized 48 American Society of Anesthesiologists (ASA) grade 1–3 patients undergoing day case cholecystectomy to liberal (40 ml/kg) or restricted (15 ml/kg) perioperative fluid therapy. Those receiving liberal therapy had fewer postoperative problems and were more likely to achieve same-day discharge [57]. She then randomized 32 ASA 1–3 patients undergoing inpatient colonic surgery and found better postoperative arterial oxygenation with restrictive therapy [58]. In the same year, she published a study that randomized 48 ASA 1–3 patients undergoing knee prosthetic surgery and found early postoperative pulmonary function to be better with liberal fluid therapy [59]. Doherty and Buggy expressed their view that a “restrictive” intraoperative fluid regimen, avoiding hypovolemia but limiting infusion to the minimum necessary, is likely to reduce complications after complex surgery. A single-centre chart review of 1242 colorectal surgeries later confirmed that greater perioperative fluid volume was independently associated with prolonged duration of recovery across a spectrum of surgical risk profiles [60]. In the contrasting clinical context of relatively low-risk patients undergoing ambulatory surgery, Doherty and Buggy are convinced that high-volume crystalloid infusion (20–30 ml/kg) reduces postoperative nausea and vomiting, dizziness, and pain. Those who advocate liberal fluid for symptom amelioration associated with ambulatory surgery, which induces only mild physiological stress response, usually do not take into account the increased risk of deep vein thrombosis [61]. Twigley and Hillman had worked at Charing Cross Hospital in London where Professor Greenhalgh’s team had conducted their thrombosis research, and so they shared the view that infusion of any intravenous fluid is of doubtful value for patients undergoing minor to moderate surgery who will be able to drink within 24 h, and may expose the patient to the risk of late thrombo-embolic complication. I was myself a junior anesthetist at Charing Cross Hospital at the time, and was taught to defer fluid therapy during minor to moderate surgery. If necessary, deferred fluid can be administered once the patient has recovered from anesthesia and is moving her legs.

Observation of a laparoscopic bariatric surgical practice over 1 year included 224 patients. Patients who received less than 1750 ml of intraoperative fluids experienced longer hospital length of stay when compared to patients who received more than 1750 ml. The best outcome was experienced by patients receiving liberal intraoperative infusion rates (more than 7 ml/kg/h), among whom only 15% had extended length of stay. Lower rates of intraoperative fluid administration were also associated with delayed wound healing [62].

Researchers in Plymouth, England, used a liberal perioperative fluid protocol (about 13 ml/kg/h) on 220 patients undergoing rectal resection or cystectomy, and randomized them to an intervention group who received additional modified fluid gelatine to achieve hemodynamic goals according to stroke volume variation. Endpoint was serious complications by Day 5, and they found no difference between the groups. They did, however, concede that the observed complication rate for patients in their study was relatively high [63]. In another study of cystectomy surgery, Wuethrich and colleagues found fewer complications, including death, in patients managed with a restrictive deferred hydration strategy, aided by a pre-emptive low-dose norepinephrine infusion [64]. Researchers in Melbourne, Australia, used a restricted perioperative fluid strategy for 100 patients undergoing colorectal surgery and randomized half their patients to additional fluid therapy guided by optimization of stroke volume index. They found no difference in postoperative complications [65].

The international multicenter randomized controlled trial called RELIEF (REstrictive versus LIbEral Fluid Therapy in Major Abdominal Surgery) succeeded in recruiting almost 3000 patients within 3 years, and was published in 2018 [66]. It is a matter of great regret that modern enhanced recovery principles were not consistently applied in the care of these patients, and that there was no Usual Care control group. Patients in the liberal fluid group received a perioperative median of 6 liters, and those in the restrictive fluid group received a median of 3.7 l. However, the resulting increase in body weight (along with fluid retention) was modest—just 1.6 kg in the liberal fluid group and 0.3 kg in the restrictive fluid group. Indeed, a quarter of the patients for whom weight data were available (my emphasis) suffered fluid deprivation with a weight loss of greater than 1 kg at 24 h after surgery. The headline conclusion was that the restrictive fluid protocol was not associated with a higher rate of disability-free survival than their liberal fluid regimen 1 year after surgery, but was associated with a higher rate of acute kidney injury.

By way of response, surgeons at the Mayo Clinic examined their own data set of more than 40,000 patients undergoing elective colorectal procedures within an enhanced recovery pathway [67]. In their hands the incidence of early acute kidney injury was much lower at 2.5% than either of the RELIEF protocols (8.6% and 5%), and long-term sequelae were exceptionally low. Of course, they may have been operating on a healthier patient population, but the finding that acute kidney injury patients received higher fluid volume on the day of surgery (3.8 ± 2.4 vs. 3.2 ± 2 L, p = 0.01) ran contrary to the RELIEF study message. The association between acute kidney injury and increased postoperative weight gain at the Mayo Clinic was even stronger at post operative day 2 (6 ± 4.9 vs. 3 ± 2.7 kg, p = 0.007), but could have been a consequence of the harm rather than its cause.

One reason for less fluid retention in this trial may be that many surgical procedures performed today are minimally invasive, which reduces the metabolic stress that leads to arginine vasopressinaemia. Another reason may be the hypotonic or balanced intravenous fluids that were administered in the current study, which were associated with a lower osmotic load than isotonic fluids.

Special Case Surgery?

In recent years there have been several observational and retrospective reports of perioperative fluid protocols to reduce specified complications of certain surgical procedures. Transsphenoidal pituitary surgery is often complicated by hyponatremia, but a retrospective review of 344 patients showed no evidence that fluid restriction or diuretic therapy was helpful [68]. Data from 65 neonates undergoing systemic-to-pulmonary artery shunt surgery showed no correlation between the time to negative fluid balance and the duration of mechanical ventilation or hospital length of stay [69]. In 365 male patients undergoing rectal cancer surgery, intra-operative fluid restriction did appear to reduce the risk of post-surgical urinary retention [70]. 40 patients undergoing lung resection were maintained on a euvolaemic (volume balanced) perioperative fluid protocol with preserved global end diastolic volume and increased cardiac output but no increase in lung water volume. Three patients (7.5%) did, however, experience acute kidney injury [71]. In 1442 lung resection patients from another centre the incidence of acute kidney injury was around 5% but appeared to be unrelated to crystalloid therapy volume. Acute kidney injury occurred more often amongst patients given hydroxyethyl starch [72]. Data from 54 patients randomized patients undergoing pancreatic surgery with lower (5 ml/kg/h) or higher (10 ml/kg/h) fluid infusion rates showed no differences in indicators of recovery of gastrointestinal function [73]. Intraoperative fluid restriction in patients undergoing pancreatic surgery has not yet been shown to affect postoperative outcome [74]. Surgeons at Duke Medical Center, USA, use a fluid restricted protocol in kidney donor surgery in the belief that restricted use of intraoperative fluids prevents excessive third spacing and bowel edema, enhancing gut recovery without adversely impacting recipient graft function [75]. Intraoperative fluid restriction has also been used in 164 Living liver donor hepatectomy cases with the intention of achieving low central venous pressure with reduced blood loss while avoiding acute kidney injury [76]. Severe intra-operative declines in renal function have been observed shortly after liver transplantation. Elevation of renal oxygen consumption occurs but is not matched by a proportional increase in renal oxygen delivery [77]. In 147 patients undergoing brain surgery, goal-directed fluid restriction protocol was associated with a reduction in intensive care unit length of stay and costs, and a decrease in postoperative morbidity [78]. Amongst 169 patients who underwent cytoreductive hyperthermic intraperitoneal chemoperfusion at the University of Massachusetts Medical School, restrictive intraoperative fluid therapy was associated with lower post operative morbidity and reduced length of hospital stay [79].

Safely Implementing a Smaller Volume Approach to Perioperative Fluid Therapy

Fluid Balance Monitoring

With general acceptance that perioperative fluid deprivation (negative fluid balance) predisposes to renal injury it is surprising that unintentional fluid deprivation is not uncommon and compliance with fluid balance monitoring in practice remains inadequate [80,81,82,83]. Quality initiatives are indicated, especially if very restrictive policies are to be implemented.

Infuse Colloids?

The whole raison d’aitre of biophysical colloid osmotic pressure therapy is to achieve elevated central volume of extracellular fluid while keeping the tissue volume low. There are many problems with this strategy, not least that it is far less effective than is widely anticipated. 2004 saw the publication of two landmark papers.

• In a large randomized clinical trial comparing albumin and saline for fluid resuscitation in a general intensive care patient population, it was observed that the volume of human serum albumin solution required to achieve resuscitation on the first day (mean 1.2 l) was only a little less than the effective volume of 0.9% sodium chloride (mean 1.6 l), and that albumin-treated patients received more red cell transfusions in the first 2 days [14].

• In a laboratory study it was shown that “colloid osmotic forces opposing filtration across nonfenestrated continuous capillaries are developed across the endothelial glycocalyx and that the oncotic pressure of interstitial fluid does not directly determine fluid balance across microvascular endothelium” [11].

In 2009, researchers in Amsterdam confirmed, by clinical experiments in both septic and nonseptic patients, that reducing colloid osmotic pressure of plasma does not predispose to pulmonary edema, and that colloid resuscitation does not reduce the risk [84]. In the hope that hydroxyethyl starch might succeed where albumin had failed, a large randomized controlled trial was published in 2012 and confirmed that the plasma substitute causes more harm than isotonic salt solution and has very little volume advantage in clinical resuscitation [15]. In 2013 it was shown that perioperative stroke volume optimization goal-directed fluid therapy is possible with crystalloid, and there is no evidence of a benefit in using hydroxyethyl starch . The authors declared that “the concept of the 1:3 replacement ratio in hypovolemic patients is obsolete” [85]. A 2016 Cochrane meta-analysis of colloids versus crystalloids in critically ill, trauma and surgical patients found 59 clinical trials involving nearly 17,000 patients showing no clinical advantage for the use of colloids. In trials recruiting patients with sepsis colloid use increased the risk of developing acute kidney injury requiring renal replacement therapy [86]. It is sometimes speculated that colloids do not harm patients undergoing uncomplicated surgery, but it is established that they confer no advantage over crystalloids. Trauma, surgery, and sepsis are all associated with systemic inflammatory response with increased capillary escape rate for albumin, and there is no reason to distinguish them for the purposes of rational fluid prescribing.

The Cochrane collaboration continues to advise against the use of colloids for volume resuscitation [87]. While some anesthetists still feel incapable of managing surgical patients without biophysical osmotic therapy, the volume kinetic fact is that the rate of elimination of crystalloid infusions is greatly retarded by general anaesthesia [88]. Crystalloids will do the job in the operating room.

The advantages that early physiologists predicted for intravenous biophysical colloid osmotic pressure therapy do not happen, and trials show more signal for harm than for benefit. A full appreciation of modern Starling physiology points to better ways to influence the intravascular—extravascular volume ratio.

Oliguria

A systematic review in 2016 examined the significance of peri-operative oliguria in randomized controlled trials. The incidence of oliguria tended to be greater with restrictive protocols, but there were too few data to achieve statistical significance. Oliguria occurs even in patients given liberal fluid volumes. Acute renal failure occurred equally in restrictive and liberal groups [89]. Enhanced recovery pathways are increasingly de-emphasising the traditional half a ml per kg rule. Acknowledging that obligatory urine volume is not proportional to body weight, I have personally suggested that oliguria in adult patients should be defined as less than 80 ml in any consecutive 4 h. In a recent study restrictive hydration with norepinephrine administration was as safe for renal function as liberal hydration intraoperatively [90]. Low-dose alpha −1 agonist infusion counteracts anesthesia-associated oliguria and may be a more effective approach than giving more fluid.

Thirst

There is an interesting hypothesis that the experience of thirst is a good indicator of need for additional intravenous fluid. A system that delivers intravenous fluid in response to subjective thirst has been designed, and can correct fluid deficits in healthy volunteers. The next step will be a feasibility study to assess this system in the perioperative setting [91].

Reduce Venous Capacitance

While euvolemic cardiovascularly fit patients should not need hemodynamic support under anesthesia that is not too deep, higher-risk patients are sometimes given a titrated dose of vasopressor to maintain vascular tone. See, for example, the protocol of Jacob et al [26]. In higher (hypertensive) doses the alpha-1 effect is predominantly precapillary vasoconstriction, while there is experimental evidence that propofol-induced hypotension is largely due to postcapillary venodilation and is more appropriately treated by restoring the mean systemic pressure and the venous excess volume. Robert Hahn has shown in a human clinical experiment that an infusion of phenylephrine 0.001 mcg/kg/h prevents anesthesia-induced hypotension by slowing distribution of infused crystalloid from the central to the peripheral fluid kinetic volume and so preserving the plasma volume [92]. Through the lens of RSE&GM, I prefer to think of the mechanism as reduction of J_v due to reduced capillary pressure. Low-dose phenylephrine infusion increases urinary excretion, counteracting anesthesia-induced oliguria, without slowing heart rate or raising blood pressure, and does not reduce the plasma volume as might occur at a higher dose. Another conceivable advantage is alpha-1-mediated increased lymphatic spontaneous contractility, which would aid return of interstitial fluid to the bloodstream. Wuethrich and colleagues have demonstrated that what they call preemptive norepinephrine infusion at 2 mcg/kg/h can be used to limit fluid balance and to improve patient outcomes after cystectomy [64].

Sodium Dose

Physiology predicts there will be very little, if any, excretion of electrolyte-free water in many hospitalized patients, because arginine vasopressin opens aquaporin-2 channels of the renal collecting ducts making them permeable to water. The clinical importance of what we might call arginine vasopressinemia is too often ignored by modern fluid guideline writers. Many of the drugs given to adults in hospital either stimulate the further release of arginine vasopressin or increase the sensitivity of arginine vasopressin receptors. Disorders of volume and tonicity are therefore the most common serious problems associated with intravenous fluid therapy. An editorial comment warned that “there can be no justification for administering hypotonic fluids in the perioperative setting” [93]. Moritz warned anesthesiologists that their most stressed trauma and surgery patients are under the endocrine influence of arginine vasopressin and are very susceptible to symptomatic reductions in plasma sodium when fluid, even an isotonic solution, is infused. He recommends 0.9% sodium chloride in glucose 5% solution for postoperative fluid maintenance, and cautions that no hypotonic infusions should be given to patients whose plasma sodium is <138 mmol/l. In neurosurgical practices hypertonic salt solutions may have to be used to keep plasma sodium well above 140 mmol/l [94]. Surgeons at Thomas Jefferson designed a study to determine whether 3% hypertonic saline could reduce the volume of fluid required to sustain tissue perfusion in the perioperative period and improve outcomes for pancreaticoduodenectomy patients. Two hundred sixty-four patients completed the study, which confirmed that perioperative hypertonic saline prescription achieves smaller net fluid balance and reduces complications [95]. Moritz and I have criticized UK guidance, which unfortunately advocates fifth normal saline in glucose for postoperative fluid therapy [96, 97]. Vulnerable patient groups are:

• Children and young adults who normally have a higher proportion of intracellular to extracellular fluid volume within the cranial cavity

• Patients with deficient blood–brain barrier including meningitis, encephalitis, trauma, tumor

• Older adults and patients with neuromuscular disorders leading to reduced muscle mass, which is a major extracellular fluid reservoir

• Premenopausal adults (estrogen)

• Trauma or surgery (arginine vasopressin )

• Concurrent drugs including morphine, nonsteroidal anti-inflammatory agents, anticonvulsants (arginine vasopressin )

• Endocrine abnormalities including hypothyroidism, adrenal insufficiency, syndrome of inappropriate antidiuretic hormone (ADH) secretion

We must dispel any notion that any one intravenous fluid is somehow superior to another, or that there can ever be a single universal resuscitation or maintenance fluid. The earliest isotonic salt solution was described by the Dutch physiologist Hartog Hamburger. Armed with Hamburger solution (0.9% sodium chloride), 1.26% sodium bicarbonate and 5% glucose solutions, plus potassium supplements as needed, the intelligent prescriber can match the fluid and electrolyte needs of almost any individual patient under his care. If it becomes necessary to infuse more than 2 l (30 ml/kg) of isotonic salt solution in a day, chloride/ bicarbonate balance and plasma acid/ base status may become significant considerations. We do not know whether hyperchloremia is more to be feared than hypochloremia, and it is quite possible that abnormal serum chloride is a symptom of the underlying disease rather a complication of fluid therapy. A Cochrane collaboration found that the administration of isotonic sodium chloride (unbuffered) or buffered fluids to adult patients during surgery are equally safe and effective [98]. The use of buffered fluids is associated with less hyperchloremia but the clinical significance of chloremia is not known. Nonetheless, the intelligent prescriber can either use both isotonic sodium chloride and isotonic sodium bicarbonate in appropriate proportions, or prefer a so-called balanced salt solution. For some reason British and Antipodean anesthetists honor the American pediatrician Hartmann and often use his solution, while Americans and Europeans honor the English physiologist Ringer and use his lactated solution. Hartmann’s and Ringers lactate are essentially the same. They are hypotonic rather than isotonic solutions and it would be dangerous to use them in the resuscitation of vasopressinemic patients who are susceptible to harm from hyponatremic encephalopathy. Consider the following calculated example of how harm can ensue. The osmolalities of plasma, 0.9% sodium chloride and Hartmann’s solution are about 288, 286, and 256 mosmol/kg, respectively. We therefore expect infusion of Hartmann’s solution to reduce plasma osmolality in a dose-dependent fashion in patients who cannot excrete a free water load. Consider what happens with just a 3% reduction in plasma/extracellular fluid osmolality from 288 to 280 mosmol/kg; there will be a 3% (40 ml) increase in intracellular brain volume, which must cause a 40 ml (30%) decrease in intracranial blood and cerebrospinal fluid volume. In a patient with critically compromised intracranial compliance, such a change can be fatal. Another consideration is that the infused lactate will make it impossible to use plasma lactate measurements as an indicator of tissue perfusion [99]. There are a number of isotonic balanced salt solutions that include anions other than lactate, and they could be a rational choice for your practice.

The restrictive fluid therapy protocol used in the University of Melbourne study is a reasonable template for patients undergoing major abdominal surgery. Carbohydrate drinks are supplied up to 2 h preoperatively. Intravenous fluid preload is not recommended, but a small postinduction bolus of up to 5 ml/ kg could be administered if hypotension is thought to be due to hypovolemia. Intraoperative balanced isotonic maintenance fluid rate was 300–400 ml/h (5 ml/kg/h), reduced after surgery to 40 ml/h (0.5 ml/kg/h for larger patients). Urine output criterion for post-operative oliguria is 120 ml per 4 h. Additional non-sanguinous fluid could be prescribed to replace blood loss, for which 2 ml crystalloid per milliliter of blood should suffice in non-major hemorrhage, or to treat hypotension that did not respond to a vasopressor. In the United states Glucose 5% sodium chloride 0.9% is often used for maintenance infusion in acutely ill and major surgical patients, but for safety reasons this solution is very rarely used in adult practice in the United Kingdom [100]. In the United Kingdom, Enhanced Recovery After Surgery (ERAS) programs are widely adopted and advocate zero-balance fluid goals for uncomplicated surgery. Postoperative fluid therapy is to be kept to a minimum and oral intake encouraged. Postoperative oliguria (<0.5 ml/kg/h) is accepted if there is no other cause [53]. It is reported that uptake of ERAS in the United States has been slow, perhaps for fear of expense. Stone et al. suggest that financial savings are in fact attainable [101].

I was a contributing member of the recent Association of Anaesthetists of Great Britain and Ireland (AAGBI) report on perioperative care for diabetic patients. By consensus we suggested half-normal saline in glucose for routine postoperative maintenance in this patient subgroup receiving insulin therapy [102]. Where it is available, normal saline in glucose may be preferable. Hartmann’s solution increases glycemia and is probably best avoided in diabetics.

Protect Lymphatic Pump Efficiency

There is a common misunderstanding that lymph is a passive overflow drainage system, when in fact it is part of a vital extracellular fluid circulation and is pumped by several mechanisms that can be optimized. Lymphatic smooth muscle responds much as cardiac muscle does to volume and neurohumoral factors. The rate of contractions and their strength are enhanced by alpha adrenergic agonists (lymphogogues) and inhibited by many inflammatory mediators, anaesthetic techniques and opioids. Intrathoracic pressure cycling moves lymph towards the central veins because lymphangions contain semilunar valves. Somatic muscle activity motivates lymph (and venous blood) flow.

Monitoring for Euvolemic Goal-directed Therapy

It is presumed that early detection of reduced stroke volume enables early correction and avoidance of inadequate tissue perfusion, which must be harmful. We can accept the premise that increasing ventricular responsiveness to cyclical changes in preload induced by mechanical positive-pressure ventilation precedes reduced stroke volume state, It has been found that, in experienced/expert hands, dynamic parameters such as arterial pulse pressure variation and stroke volume variation are reasonably accurate predictors of fluid responsiveness. There are, however, three important exceptions: (1) when tidal volumes are low or the patient is breathing spontaneously, (2) when the chest is open, and (3) during sustained cardiac arrhythmia. The utility of this strategy is to achieve zero-balance fluid management while avoiding hypovolemic circulatory compromise. Unfortunately, in the hands of practicing anesthetists, an observational study of the randomized controlled trial OPTIMISE found that the predictive accuracy of stroke volume variation and pulse pressure variation for fluid responsiveness was not adequate for routine use during or after major gastrointestinal surgery. The data confirmed the established view that these variables should not be used for predicting fluid responsiveness in spontaneously breathing patients [103].

Brandstrup randomized 150 patients undergoing colorectal surgery to a zero-balance protocol or an optimal stroke volume protocol and found no difference in outcome [104]. Her fluid management involved hydroxyethyl starch in both treatment limbs. A Cochrane systematic review published in 2013 identified 31 studies involving 5292 patients. It highlighted the paucity of recent data concerning goal-directed therapy; Sandham’s trial published in the New England Journal of Medicine a decade earlier, using pulmonary artery catheter monitoring, dominated the data set. The reviewers found that perioperative cardiac output monitoring did not reduce perioperative mortality, but might reduce the number of nonfatal renal and pulmonary complications and poor wound healing. and reduce the overall length of hospital stay by about 1 day. Rates of cardiac arrhythmia, myocardial infarction, congestive cardiac failure, venous thrombosis, and other types of infections were not reduced. They opined that the evidence did not support widespread use of perioperative cardiac output monitoring [105]. Later trials have supported their conclusion. Unsurprisingly, arterial pulse contour analysis did not improve outcomes for elderly spontaneously breathing patients under spinal anesthesia [106]. OPTIMISE enrolled 734 high-risk patients undergoing major gastrointestinal surgery and could not demonstrate an advantage for hyperdynamic therapy [107]. By the sophistry of adding their data to a meta-analysis, they were able to claim a statistically significant reduction in the complication rate. The OPTIMISE II trial is hoping to complete recruitment of 2500 patients by the end of 2019 (https://optimiseii.org/about}. On the current evidence, it is my view that stroke volume monitors that can be used pre- and postsurgery can be a valuable part of the expert anesthetist’s intraoperative armamentarium to attain and maintain euvolemia, to be used on his/her clinical judgment. Rigid protocols that are allowed to override informed clinical judgment may be dangerous.

Monitoring for Hypervolemic/Hyperdynamic Goal-directed Fluid Therapy

Since the pioneering work of William Shoemaker in California, it has been hoped that supranormal stroke volume achieves supranormal oxygen delivery, which aids healing and reduces complications of surgery [108, 109]. The premise in this strategy is the same as for euvolemic GDFT, but the utility is to achieve a hyperdynamic state with least harm from edema. In Intensive Care practice, hyperdynamic therapy guided by cardiac output monitoring has been found to be ineffective at improving patient outcomes in three major studies [110,111,112]. A more recent study of 187 patients undergoing major surgery confirmed that patients achieving higher oxygen delivery experience fewer complications, but that a protocol specifically targeting higher oxygen delivery does not improve surgical outcome [113]. In my view, this strategy is physiologically unsound as well as demonstrably ineffective and should no longer be pursued.

Avoid Fluid Boluses

One must exercise great caution in extrapolating fluid kinetics of healthy volunteers to patients. Nonetheless, Hahn’s demonstration by volume kinetics that edema is a normal consequence of plasma volume expansion in healthy volunteers when crystalloid fluid is given rapidly, but the tendency for edema formation is small when the fluid is given slowly, is compelling [24]. While Hahn reasons that this is largely due to visco-elastic properties of the interstitial matrix , I feel that transendothelial hyperfiltration (excessive J_v) during transient peaks of capillary pressure is equally plausible. In an intra-operative patient study, neither colloid nor crystalloid boluses increased oxygen delivery, and only crystalloid boluses increase the glomerular filtration rate. Renal oxygen consumption rose along with the increase in glomerular filtration but was not matched by a proportionate increase in renal oxygen delivery, raising concern about adequate renal oxygenation [114]. Such evidence provides physiological reasons why bolus therapy was observed to be harmful in the Fluid Expansion As Supportive Therapy (FEAST) trial [16]. It is my view that research is urgently needed on the role of bolus fluid therapy. Until more evidence is available, clinicians should use smaller boluses at lower infusion rates, repeated as needed, to achieve the desired central fluid volume expansion.